Advancing privacy-preserving AI skin analysis: Dermaself and Alta Scuola Politecnica
A smartphone selfie contains two different types of information: visible characteristics of the skin and biometric features that can identify the person in the image.
For beauty tech and digital dermatology, being able to work with these two layers independently opens new opportunities. Skin information can remain valuable for analysis and research, while identity can be protected through dedicated privacy-preserving technologies.
This was the focus of a research project developed by a multidisciplinary team from Alta Scuola Politecnica, the joint honours programme of Politecnico di Milano and Politecnico di Torino, in collaboration with Dermaself.
The team worked on two complementary technologies designed to further enhance selfie-based AI skin analysis: image normalisation and 3D-based facial anonymisation.
One research team across Politecnico di Milano and Politecnico di Torino
The project brought together five master's students from the two universities behind Alta Scuola Politecnica.
Andrea Germano and Adriano Giuliani, studying Computer Engineering, and Federico Greppi, studying Biomedical Engineering, represented Politecnico di Torino.
Edoardo Gribaldo, studying Computer Science and Engineering, and Alessia Soccionovo, studying Management Engineering, represented Politecnico di Milano.
The project was guided by academic tutors Professor Elisabetta Raguseo of Politecnico di Torino, Director of Alta Scuola Politecnica, and Professor Federica Arrigoni of Politecnico di Milano.
Working directly with Dermaself, the team explored how advanced academic research in computer vision, image processing and privacy could be translated into technologies with real-world applications in beauty tech.
Creating increasingly standardised inputs for AI skin analysis
The first area of research focused on selfie image normalisation.
Images captured with smartphones naturally vary according to factors such as lighting conditions, exposure and camera processing. The project therefore explored an additional image-processing pipeline designed to further standardise photographs before they are processed by AI systems.
During image acquisition, users are guided towards suitable conditions, including appropriate positioning, distance and lighting. After capture, the technology can further compensate for variations in illumination through a hybrid pipeline combining AI-driven illumination correction with classical image-processing algorithms.
The result is a more standardised representation of the image, designed to make selfies captured across different devices and environments increasingly comparable.
For Dermaself, this research represents an additional technological layer that can complement its existing AI skin analysis technology and support increasingly consistent image acquisition across multiple customer touchpoints.
Preserving skin detail while protecting identity
The second area of research explored a more ambitious question: whether the information contained in a facial image could be separated from the identity of the individual represented.
Traditional anonymisation techniques, such as blurring or masking, can remove identifying features but may also eliminate valuable visual information from the skin.
The Alta Scuola Politecnica team therefore developed a different approach based on 3D facial reconstruction.
The research explores a new paradigm for privacy-preserving facial imaging, moving beyond conventional anonymisation techniques such as masking or blurring. Using a proprietary 3D computational framework, the technology separates identity-bearing facial information from the high-fidelity visual signals required for skin assessment.
The resulting transformation operates at the structural level of the facial representation, substantially altering its biometric signature while retaining clinically meaningful skin characteristics, including lesions, redness, pigmentation, spots and fine-grained texture.
The key breakthrough lies in treating identity and skin as computationally distinct information layers within the same image. This makes it possible to preserve the richness of the skin data rather than destroying it in the process of protecting the individual’s identity.
From academic research to Dermaself's AI skin analysis ecosystem
The technologies developed through the Alta Scuola Politecnica project have been conceived to complement Dermaself's AI skin analysis ecosystem and expand its technological capabilities.
Dermaself enables beauty brands and retailers to integrate selfie-based skin analysis into digital and physical customer experiences. From a selfie, its technology analyses visible skin characteristics and connects the results with products from a brand's or retailer's own catalogue to create personalised skincare recommendations.
The platform can be integrated across different touchpoints, including e-commerce platforms, beauty retail environments and brand events.
Within this ecosystem, image normalisation can provide an additional layer of standardisation across acquisition environments, while the anonymisation technology opens new possibilities for privacy-preserving image processing.
The research also creates opportunities beyond individual skincare recommendations, particularly in areas where large volumes of skin images could support research, technology development and AI model improvement.
New possibilities for privacy-preserving beauty datasets
High-quality image datasets are increasingly valuable for companies and research organisations developing computer vision and AI technologies for skin analysis.
Facial images, however, require particular attention because they contain biometric information alongside the visual characteristics being studied.
The ability to preserve relevant skin information while reducing identifiable facial characteristics could enable new approaches to building and managing privacy-conscious image datasets.
In the future, this could support research collaborations, technology development and new data-driven services across beauty tech and digital dermatology, subject to the appropriate privacy, legal and validation frameworks.
For Dermaself, the project represents another step in the development of an AI skin analysis ecosystem in which technological performance and privacy can evolve together.
By combining Dermaself's experience in applied AI skin analysis with the research expertise of Alta Scuola Politecnica, Politecnico di Milano and Politecnico di Torino, the collaboration demonstrates how academic research can be translated into technologies designed for real-world beauty applications.
Frequently Asked Questions
What is AI skin analysis?
AI skin analysis uses artificial intelligence and computer vision to analyse visible skin characteristics from an image, such as a smartphone selfie. The resulting information can be used to support personalised skincare experiences and product recommendations.
How does Dermaself's AI skin analysis work?
Dermaself analyses visible skin characteristics from a selfie and connects the results with product and ingredient information from a brand's or retailer's own catalogue. This enables brands to provide personalised skincare recommendations through digital and physical customer touchpoints.
Why does selfie lighting matter for AI skin analysis?
Lighting, exposure and camera processing can influence how skin appears in any photograph. Image normalisation is designed to reduce these variations and create increasingly standardised inputs for computer vision systems, particularly when images are captured across different devices and environments.
What did Alta Scuola Politecnica develop with Dermaself?
The project focused on two complementary technologies: a selfie image normalisation pipeline and a 3D-based facial anonymisation system. Both were developed to expand the technological capabilities surrounding selfie-based AI skin analysis.
How can a selfie be anonymised while preserving skin information?
The technology developed by the Alta Scuola Politecnica team leverages advanced 3D facial modelling to fundamentally transform the biometric signature of a face without sacrificing the richness of its skin data.
Through a proprietary computational process, identity-bearing facial information is altered at a structural level, while high-fidelity skin characteristics are preserved for potential clinical and research applications. The result is a new approach to facial anonymisation that protects identity without simply removing or degrading the information that makes the image scientifically valuable.
Who developed the technology?
The research was developed by a multidisciplinary team of five master's students from Alta Scuola Politecnica, the joint honours programme of Politecnico di Milano and Politecnico di Torino, in collaboration with Dermaself and under the guidance of academic tutors from both universities.
What is selfie image normalisation?
Selfie image normalisation is the process of reducing variations in photographs caused by factors such as illumination, exposure and camera processing. The technology developed within the project provides an additional standardisation layer designed to make images captured in different environments increasingly comparable.
Why is privacy important in AI skin analysis?
A facial image contains both visible information about the skin and biometric characteristics related to a person's identity. Privacy-preserving image technologies explore ways of retaining useful visual information while reducing identifiable facial characteristics.
Could this technology be used for AI research datasets?
The anonymisation technology opens potential applications in the creation and management of privacy-preserving image datasets for beauty tech and digital dermatology research.
By separating skin information from facial identity, this approach could support future research, AI development and data-driven applications while placing greater emphasis on the protection of the individuals represented in the images.
